Fermentation Basics: Salt, Temperature, and Air

Fermentation is rigging a competition

Every ingredient carries a mixed population of microbes. Fermentation makes conditions favour the lactic acid bacteria so they crowd the rest out. They convert sugar to lactic acid, the pH falls, and spoilage organisms cannot survive a low pH.

So the safety of a ferment does not come from cleanliness — it comes from the acidity dropping fast. Which is why the first few days matter most.

Where the acid comes from decides whether there is a window at all

Everything above assumes the acid has to be made in the jar — the pot starts near neutral, the microbes have days of work to do, and the outcome is in doubt while they do it. That is what makes the first few days the risky ones. Where the acid is in the jar from the first minute there is no competition to rig and no window to survive, and the levers below stop being what governs. Vinegar appears further down this page once, as something that was itself a ferment — never as a way of arriving at the same pH without one.

It can be poured in. Pikliz says it plainly — "the acid is brought in, not made" — and then draws the consequence for the lever immediately below: "5% vinegar is already below the pH at which anything can live, so there is no need for salt to select the microbes. The teaspoon of salt is for flavour." Salt is not being set to a concentration there; it is out of a job, which is the exact inverse of Kkakdugi's "do not skip the salting, it is the whole dish". What the months then buy is flavour, and pikliz says what kind: in a liquid with almost no fat in it the capsaicin comes out of the chillies very slowly, so a day-old jar is mild and a month-old jar is hot. And a bought acid can be taken back out once it has done its work — Nata de coco pours vinegar in to get the bacteria working and then boils the sourness out of the finished cubes, which is not available for acid a ferment built out of the food's own sugar.

Or it can be inherited. Paocai is the case where the jar rather than the batch is the thing being made: the vegetables are changed and the same brine is kept for years, and "a jar whose lactic population is established ferments faster and more predictably — which means the first batch is the worst one you will make." So "the first few days matter most" is true exactly once, of the first batch, and after that the window has been paid for permanently. Attiéké does the same thing without the liquid and gives the figure: grated cassava left to wild flora takes three or four days the first time, and holding 200g of that batch back makes every later one two days. What protects the one exposed batch is not on the three levers at all — paocai's baijiu is "a safeguard, not a flavour. The alcohol suppresses the wrong microbes early on" (cooking with alcohol).

Nam khao is the control, and it is this premise at full strength: raw pork made safe by acid rather than by heat, the acid made inside the parcel by the same competition, nothing bought and nothing inherited. Its instruction is what the premise looks like in a kitchen — wait until it is properly sour, or cook it.

So this is not another lever. The three below are how you win a competition, and in the first two cases there was no competition to win, because the pH the microbes were being asked to produce was in the jar before the vegetables were. What changes is what an early taste tells you: in a lactic ferment a jar that is not yet sour is a jar still at risk, and in a bottled-acid jar it is only a jar that is not yet flavoured. The acid still has to be the right one — Balancing flavour carries pikliz's own note that citric is weaker than acetic, "so it keeps less well and has to stay in the fridge."

Lever one: salt concentration

Salt sets a level lactic bacteria tolerate and most spoilage organisms do not.

• 2–3% — the usual range for Kimchi and its relatives. Ferments fast, gets properly sour.
• 5%+ — for long storage. Slow, and noticeably salty.
• Under 2% — things go soft and the risk of the wrong microbes winning rises sharply.

Salting the cabbage is the step where you set this concentration. How much you rinse afterwards changes the final figure, so taste a piece after rinsing.

Lever two: temperature

Temperature sets the speed — and shifts the flavour with it.

• 20–25°C — sour within two or three days. It is why kimchi ferments overnight in summer.
• 10–15°C — one to two weeks, with sourness and savouriness developing in balance. The best window.
• Below 4°C — essentially stopped. This is where you hold a ferment you like.

So the method is always the same: ferment at room temperature to the point you want, then move it to the fridge. A kimchi fridge exists to hold that middle band.

When the fridge is where the work happens — a yeast dough

The band above describes 4°C as the place you keep a ferment that is already where you want it. A yeast dough goes in unfinished and comes out changed rather than paused. Focaccia writes down why: "twelve to eighteen hours cold gives the yeast time to produce a far wider range of aroma compounds." Gas and aroma are not on the same clock, and cold separates them — the hours buy flavour without the dough inflating past what it can hold up, which is why Brioche says "do not wait for doubling on the first prove … stop at about half again and move it to the fridge."

And in two of them the cold has nothing to do with flavour at all. It is working on the dough rather than on the ferment.

• It makes a skin. Bagels: "the overnight cold proof is not only about flavour. Without the thin skin it forms, the dough loosens in the boiling water and the surface turns ragged."
• It firms it. Brioche draws the pair itself — for a bagel the cold forms "a skin so the dough does not loosen in the boiling water", and here "it firms a dough that is otherwise far too soft to shape." In a laminated dough the same fridge is holding the butter solid at the same time, as Pain au chocolat does (dough and gluten).

So unlike a ferment you are holding, the dough comes back out. Kanelbulle gives it thirty minutes at room temperature before baking and pain au chocolat an hour: it has to be returned to the temperature the yeast works at before it meets the oven, where a kimchi is served straight from the cold. Liège waffles move the whole first rise into the fridge.

The breads this guide does name go the other way — salt-rising bread is held at 38-45°C and jeungpyeon hands the job to makgeolli yeast. Taking the temperature down was the direction with no entry.

Lever three: keeping air out

Lactic bacteria do not need oxygen; moulds do. So pressing everything under the liquid is the whole anti-mould strategy.

• Pack it down so the brine covers the solids.
• Cover the surface with an outer leaf or a sheet of film to limit air contact.
• Do not fill the jar — leave about 20% headspace, because gas is produced. A sealed jar filled to the brim overflows.

Normal versus failed

• A thin white film on the surface → kefir/kahm yeast. Skim it. If it smells sour, it is fine.
• Fuzzy black, green or pink mould → discard it. Scraping the visible part off does not help; the hyphae run deeper. (Tempeh, which grows a mould on purpose, is the exception — see the section below.)
• Bubbling and a sour smell → normal. It is working.
• Slimy, with a rotten smell → failed. Not enough salt, or too warm.
• Brine gone thick and stringy → wrong microbes. Discard. (Natto is the exception: there, stringy is the finished state.)

When the levers run backwards

The three levers above assume a lactic ferment. A ferment that runs on something else turns the same levers the other way — so a reader who knows only the rules above will throw away a finished dish.

• No salt, at 40°C, with air let in — the Bacillus in Natto and Cheonggukjang needs oxygen and is poor at tolerating salt, and the pH rises instead of falling. Safety there comes from a pure starter and a scalded container, not from acidity.
• A mould, grown deliberately, in air — the koji in Miso is grown on a tray in 30°C air, the exact opposite of lever three. And that mould does not ferment anything: it manufactures enzymes and then stands aside.
• Hyphae running deep is success — in Tempeh the hyphae growing between one bean and the next are the binder. "The hyphae run deeper" from the failure list above is the objective here.
• No liquid to submerge anything under — Lahpet presses steamed tea leaves under a stone and the juice they give up becomes the brine. Takuan is buried in a dry rice-bran bed with no liquid at all.
• One lever's range multiplied by four — Umeboshi is 20% salt. At that concentration neither lactic bacteria nor mould grows: it is a pickle in which nothing grows, not a ferment.
• Above every temperature band — Salt-rising bread is held at 38–45°C, the band where yeast and lactic bacteria lose. That is how it wins the competition with no starter at all.
• Acid as by-product, not aim — Jeungpyeon rises on the CO2 from makgeolli yeast and the lactic acid is what comes along with it. Injera is also a wild ferment, but there the acid is the flavour.
• A different reaction entirely — in Hongeo-hoe urea becomes ammonia and the pH climbs to about 9. Haring is autolysis, the herring's own pancreatic enzymes cutting the flesh, and salt there is a ceiling rather than a floor. Nata de coco does not transform its substrate at all — the bacteria excrete cellulose, and what you eat is that.

Ferments as ingredients

Fermenting is cooking in itself, but most often it is an ingredient for something else.

• Kimchi is three different ingredients: fresh, ripe, and very sour. Very sour kimchi suits Kimchi Bokkeumbap and kimchi stew because the acid carries fat and meat.
• The doenjang in Doenjang Jjigae and the miso in Miso Soup are soybeans fermented with mould and yeast, and boiling kills both the aroma and the live cultures. That is why they go in at the end and are not boiled hard. What that mould actually does is on the Miso page.
• Dengaku puts that same miso straight over charcoal. The ferment's own aroma is driven off and its free amino acids brown into a different one — not a rule broken so much as a different thing wanted.
• Some ferments make an ingredient. Attiéké inoculates fresh cassava with a lump of the previous batch. Nata de coco has you eat the material the bacteria built.
• Soy sauce, fish sauce and vinegar are all ferments too. More on that in Building umami.

The one to start with

The hardest ferment to get wrong is brined vegetables. One litre of water, 20–25g salt, radish or cucumber or cabbage held under the surface, two or three days at room temperature. When it bubbles and tastes sour, refrigerate. That single batch teaches all three levers.